Training machine learning systems for seismic interpretation

This invention describes a way to teach an artificial intelligence system to interpret underground seismic images, which are used to find oil and gas. It works by training two AI models: one model learns to interpret the images, while a second model acts as a critic, ensuring the interpretations are geologically realistic. The claims specifically narrow this to an adversarial training method where the interpreter model tries to generate interpretations that fool the critic model, while also ensuring the interpretations fit the original data and known geological rules.

Why it matters: Since 2019, the stability and performance of adversarial training methods, like those described, have significantly improved. This makes the complex task of integrating geological priors into machine learning models more robust and efficient to implement today.

Status: Not in forceListed as no longer active. The specific reason is not in the record we hold.
How hard to build: SpecializedHeavy software system, ML expertise

3 AI reinvention directions for this patent

See all three

AI gives you a few directions you could take this. Pick one, and we check whether your version is different enough to patent, then write the filing.

Reinvent this with AI

Where this one came from

Every morning we go through millions of abandoned patents and pick out 15 worth your time. This was one of today’s.

This one

+10